AplicaÃÃo de sistemas imunolÃgicos artificiais para prediÃÃo da estrutura de proteÃnas / Artificial immune systems aplication for protein structure prediction

AUTOR(ES)
DATA DE PUBLICAÇÃO

2007

RESUMO

This work presents hybrid Immune-based Systems to solve the Protein Folding Problem for the three-dimensional hydrofobic-Polar model (3D HP). The Protein Structure Prediction consists of finding the spacial arrangement of a proteinâs amino acids that has a minimal energy. The proposed methodology is focused on Artificial Immune Systems supported by but uses Fuzzy Inference Systems and Tabu Search. Two kinds of algorithms are applied: the Clonalg algorithm and the Immune Network algorithm. In both cases penalty based approaches are compared with approaches that allows only feasible antibodies. In some of the implemented models, an aging operator (that can be fuzzy or pure) is used to decide which antibodies will be eliminated of the population before the selection stage. Moreover, two affinity maturation stages were implemented: one weak and one intensive. The first is based on the hypermutation and hypermacromutation operators while the second is based on the tabu search metaheuristic. To validate the results, the models were applied to some instances of the Tortilla benchmark. The results show that the models based on the Clonalg algorithm are superior to the models based on the immune network theory. The best results were achieved with the model that used the three technics previously commented. These results are comparable to those reported in the literature. In some cases new energy values were found by the proposed methodology

ASSUNTO(S)

immune system - artificial intelligence engenharia biomÃdica sistema imunolÃgico - inteligÃncia artificial ciencia da computacao biomedical engineering sistemas difusos diffuse systems

ACESSO AO ARTIGO

Documentos Relacionados